Challenges in Computing Electron-Transfer Energies of DNA Repair

Subscriber access provided by University of Oklahoma

Article

Challenges in computing electron-transfer energies of DNA repair using hybrid QM/MM models Abdul Rehaman Moughal Shahi, and Tatiana Domratcheva J. Chem. Theory Comput., Just Accepted Manuscript • DOI: 10.1021/ct400537b • Publication Date (Web): 26 Aug 2013 Downloaded from http://pubs.acs.org on August 31, 2013

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Chemical Theory and Computation is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Chemical Theory and Computation

1

Challenges in computing electron-transfer energies of DNA repair using hybrid QM/MM models Abdul Rehaman Moughal Shahi and Tatiana Domratcheva* Department of Biomolecular Mechanisms, Max Planck Institute for Medical Research, Jahnstrasse 29, 69120, Heidelberg, Germany e-mail: [email protected] The influence of the molecular environment on chemical activity is an important factor in biomolecular mechanisms. We studied the effects of ionic groups, that is a protonated histidine side chain and deprotonated phosphates of DNA, on electron transfer in light-induced DNA repair. On the basis of the X-ray crystal structure, we prepared a hybrid QM/MM model of the macromolecular complex formed between the (6-4) photolyase enzyme and the DNA substrate containing the thymine-thymine (6-4) photoproduct. At the optimized geometries, we computed with the CASSCF and CASPT2 methods the excited states of the electron donor and electron acceptor complex, consisting of the reduced flavin and the (6-4) photoproduct. The donoracceptor complex interacts with its environment comprised of the protein, the double-stranded DNA substrate with its counterions and the solvating water molecules, which we modeled using the AMBER94 force field. The excited states of our interest include two locally-excited (LE) states of the flavin chromophore and intermolecular electron-transfer (ET) states. We identify only minor changes of the LE excitation energies by interactions with the environment but in

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2

drastic contrast to that, we found significant changes of the ET excitation energies. In the presence of the positively charged His365H+, the ET excitation energies decrease, indicating facilitated electron transfer. In addition, the excitation energy of second LE state, explaining the flavin’s absorption at 380 nm, undergoes a 0.2-eV downshift. In contrast to the active-site protonation, reduced screening of the DNA phosphates increases the ET excitation energies but not the LE excitation energies. Accordingly, the electron affinities of the (6-4) photoproduct are significantly reduced which should hinder electron transfer from the excited flavin. We also show that dynamic electron correlation accounted by the second order perturbation theory CASPT2 does not alter the energy trends obtained with the CASSCF method. Including the histidine side chain in the QM part enhances the effect of the histidine protonation on the ET energies. We also note that protonated His365H+ can serve as an electron acceptor. Introduction Calculations of the excited electronic states of biomolecules using combined QM/MM approaches, involving application of high-level quantum-chemical methods coupled with the classical-mechanical description of the molecular environment, belong to the frontier of computational chemistry. Over past several years, such computational studies accumulated evidence that ionic groups in biological cofactors, substrates or enzymatic active sites play an important role in the excited state evolution. It is not seldom that the charged (protonated or deprotonated) side chains are found in the vicinity of the biological chromophores bound to their protein scaffolds. The well-known retinal and GFP chromophores undergo protonation inside the protein, which changes their UV-vis spectra and photodynamics implicated in the biological function. On the other hand, anionic groups such as phosphates are also very common in biomolecules. For instance, phosphodiester groups of DNA are deprotonated and thereby


Page 2 of 30

Page 3 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3

negatively charged at the physiological proton concentration. How this negative charge impacts chemical properties, in particular electron-acceptor properties implicated in the DNA photoinduced damage and repair remains elusive. In this study, we characterize a macromolecular complex between a double-stranded DNA molecule containing the (6-4) photoproduct of the two thymine bases and the DNA-repair enzyme, the (6-4) photolyase. The (6-4) photolyase is a light-dependent redox enzyme that specifically repairs the pyrimidine-pyrimidone (6-4) photoproduct using electron-donating properties of the flavin adenine dinucleotide (FAD) cofactor1-3. The DNA repair is triggered by light absorption of an antenna chromophore, folate or deazaflavin, bound to the enzyme. Then, the excitation energy is transferred from the antenna to the reduced FADH- cofactor, which in turn donates an electron to the DNA photoproduct. The formed radical undergoes a repair reaction, which formally consists of formation of a covalent bond between the O4′H-group and the 3′T base and cleavage of the covalent bond between the 5′T and 3′T nucleobases (Scheme 1). The repair is completed by return electron transfer from the repaired DNA to the FADH cofactor. In this complex process, redox potentials of the DNA photoproduct, repaired DNA and flavin cofactor play a central role and should fall into a certain range to enable the complete electron transfer cycle. Whether this critical redox-potential balance is defined by the intrinsic properties of these molecular entities or results from their interactions with the environment is still a subject of investigation. 5'T 4' O N 6 N 4' HN OH O

3'T

N O

5'T O

3'T N 6

N 4' HO

HN

4' N

O

O

Scheme 1. The (6-4) photoproduct and two “repaired” N1-methyl-thymine molecules.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

4

Along with the electron transfer cycle, a proton transfer cycle from the conserved protonated residue His365H+ to the (6-4) photoproduct and back has been proposed to take place6. Conserved His365 (in this study we consider the Drosophila melanogaster enzyme) was suggested to be protonated even under alkaline conditions, at which the (6-4) photolyase shows higher repair activity in vitro4,5. According to the recently proposed repair scheme, His365H+ protonates the (6-4) photoproduct after photoinduced electron transfer, which opens the repair channel6. Consistent with this hypothesis, only a futile electron transfer cycle was observed in a series of His365-mutated proteins6. On the other hand, contrary to the proposed “catalytic” DNA protonation, the efficiency of the enzymatic repair is reduced almost two-times upon increasing the proton concentration from pH=9 to pH=74. The protonation state of His365 under physiological conditions is unknown. Determination of the (6-4) photolyase-DNA molecular structure by protein crystallography7 stimulated computational investigations of the DNA repair mechanism8-15. An insight was acquired into electronic properties and chemical reactivity of the (6-4) photoproduct radical anion9, its repair reaction involving protonated histidine12,14, radical recombination via backelectron transfer14, ENDOR spectral signatures of the flavin radical13, electronic excitation energies9-11, and the protonation states of the repair active site11,13. At the same time, the role of the repair-site environment in tuning the redox potentials has not been studied so far. The repair molecular mechanism remains unresolved partly because of the uncertainty concerning the protonation state of the conserved His365. While some computational studies questioned the experimental assignment arguing that His365 is rather neutral11,13, the other studies, relying on the experimental assignments, proposed several DNA repair mechanisms involving protonated histidine12,14,15.


Page 4 of 30

Page 5 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


5

In our previous computational studies we considered the electron-acceptor properties of the (6-4) photoproduct9,11. We found that the (6-4) photoproduct consists of two electron-acceptor moieties – the pyrimidone 3′base with the higher electron affinity and the pyrimidine 5′base with the lower electron affinity11. Accordingly, the two ET states associated with these two electron acceptors have rather different excitation energies in the enzymatic repair site11. Proposed protonation of the 3′base further significantly increases the electron affinity. Considering the (64) photoproduct of the two methyl-thymine bases (without the sugar phosphate backbone), we found that electron transfer coupled with the protonation of the photoproduct decreases the energy of the active site by more than 35 kcal/mol in the ground electronic state11. Such an electron transfer reaction in the electronic ground state (without photoexcitation) should deactivate the enzyme, which contradicts the current interpretation of the experimental observations. Whether this ground-state electron transfer is hindered in the macromolecular DNA-repair complex is still unknown. In the current study, we present hybrid QM/MM models of the repair complex to provide an answer to this question. We study how interactions with the specific molecular environment influence the excitation energies of the repair site. We included the electron donor and the electron acceptor in the quantum-mechanical (QM) part, for which we carried out excited-state calculations with the multi-configurational CASSCF-CASPT2 method. The interactions with the molecular mechanical (MM) part modeled with the AMBER94 force-field parameters are accounted with the electrostatic potential fitted (ESPF) operator16. We are interested in how the excitation energies of the repair site change depending on the protonation state of the conserved histidine, the distribution of the DNA counterions and the presence of the water solvent. We also study the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

6

effects of dynamic electron correlation and the inclusion of the histidine side chain in the QM part. Molecular models and computational details QM/MM computational scheme. We utilized the QM/MM electrostatic embedding approach with the ESPF operator16 implemented on the basis of the MOLCAS17,18 quantum-chemistry package and the Tinker19 molecular-mechanics package. We carried out the state-averaged CASSCF20 and the multistate (MS)CASPT221,22 calculations with the 6-31G* basis set23. To model the environment of the active site within the MM part we used the AMBER94 force field24 in combination with the TIP3P water parameters25. Molecular model of the protein and DNA complex. We prepared two QM/MM models containing either protonated or neutral His365 side chain – the HisH+ model and the His model, respectively. We used the coordinates of the crystal structure of the (6−4) photolyase from Drosophila melanogaster bound to a double-stranded DNA containing the T(6-4)T photoproduct of two thymines (PDB code 3CVU)7. After inspecting the local environment, the following protonation states of the amino-acid side chains were assigned: His19 – protonated (to satisfy a salt bridge with Glu224), all other His – the H-Nε form; all Glu and Asp – deprotonated; all Lys and Arg – protonated. The phosphates of the FADH- cofactor and of the DNA backbone were deprotonated. All crystal-structure water molecules were included in our models. The hydrogen atoms were added by using the Tinker program. To neutralize the overall electrostatic charge of the DNA-protein complex, 15 and 16 Na+ counterions were added to the HisH+ and His model, respectively, along the DNA backbone. A water shell of radius 50 Å consisting of 6000 water molecules was added using the package Packmol26. The coordinates of the water-shell molecules were optimized, keeping the coordinates of the protein and DNA constrained. After that, the


Page 6 of 30

Page 7 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


7

energy minimization of the entire models was carried out keeping constrained the coordinates of the FADH- chromophore and the (6−4) photoproduct. The energy minimization was performed with the AMBER94 force field employing the Tinker program. The force field parameters and the atomic charges for the FADH- chromophore and the T(6-4)T photoproduct were taken from the previous studies27,28. The two obtained minimum-energy structures were used as starting structures in the QM/MM geometry optimization. QM/MM geometry optimization. We performed QM/MM geometry optimization with the lumiflavin part (LFH-) of the FADH- cofactor and the T(6-4)T photoproduct (5′T base and 3′T base) in the QM part (62 atoms in total). The QM/MM boundary was accounted via hydrogen link-atoms29. To avoid overpolarization of the QM part, the charges of the MM-boundary atoms were reassigned using the charge-shift scheme30. Three link-atoms were introduced: one at the C1′-C2′ bond connecting lumiflavin to the ribityl chain and two at the two N1-C1′ bonds connecting the thymine bases to the deoxyribose moieties. The geometry optimization in the ground electronic state was performed using the CASSCF(2,2)2 method. In the CASSCF(m,n)k notation, m and n indicate the number of electrons and molecular orbitals (MOs) in the active space, respectively, and k indicates the number of electronic states. In the CASSCF(2,2)2 calculations, the active space includes the highest occupied MO (HOMO) of flavin and the lowest unoccupied MO (LUMO) of the 3′base. The geometry of the MM subsystem surrounding the QM part consisting of the amino acid residues His365, His369, Tyr423, Gln299, Asn406, water molecule W697, the ribityl and adenine fragments of the FADH- cofactor, the two deoxyribose fragments and the phosphate of the photoproduct, was optimized at every QMoptimization step using the microiterations technique31. The coordinates of the other atoms of the MM part were constrained. In the geometry optimization course, the ESPF charges were



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

8

computed with the state-averaged CASSCF electron density as implemented in the MOLCAS version 7.6. The two energy-minimized structures so obtained, the HisH+ model and the His model, were used for single-point energy calculations. Modification of the MM part. To study how the DNA negative charge influences the excitation energies, we considered several modifications of the MM part. In our QM/MM models, the double-stranded DNA contains 28 negatively charged phosphates included in the MM part. Their negative charge is screened by interactions with the protein, water molecules and Na+ counterions. In the initial protein-DNA model, we placed, depending on the protonation state of His365, 15 or 16 Na+ counterions along the DNA sugar-phosphate backbone. We identify the counterions by their distances to the phosphorous atom in the phosphodiester group of the (6-4) photoproduct. Because of the similar initial structures, the distributions of the Na+ ions in our HisH+ and His QM/MM-optimized models are very similar. To decrease the screening of the DNA negative charge, we moved one or two ions from their position that is proximal to DNA to a remote position of about 30 Å. We refer to the model with the modified MM part by indicating the displaced Na+ ion(s), e.g. MM-2.7 or MM-(2.7+7.7). In the case of the HisH+ model, we also quantify the effect of the water solvent by removing the entire water shell; this model is designated as MM-W. We did not reoptimize the geometry of the modified MM part; this simplified computational procedure satisfies our main goal to compare the change in the excitation energies of the LE and ET states with the different electronic structure. Excitation energies. We computed excitation energies of the repair active site using several approaches: by varying the content of the QM part, by introducing the described modifications of the MM part and also by considering different number of electronic states. The complete list of our models with the description of the CASSCF calculations is given in the Supporting


Page 8 of 30

Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


9

Information Table S1. First, we considered the (D,A) models that contain the electron donor and the electron acceptor in the QM part (like we did in the QM/MM geometry optimization). Then, we extended the QM part to (D,A,H) by including the histidine side chain. A hydrogen link-atom was incorporated in the Cα−Cβ bond of His365. The size of the QM part is increased to 79 and 78 atoms with HisH+ and His, respectively. The histidine side-chain geometry was optimized using the force-field parameters and not quantum-mechanically. To compute the excitation energies of the (D,A) and (D,A,H) QM/MM models, the following CASSCF calculations were carried out. First, we consider only one excited state which corresponds to the ET of the 3′base computed with the CASSCF(2,2)2 method. In the case of the HisH+(D,A,H) model with the modified MM part MM-6.9 and MM-(2.7+7.7), the ET-3′T state was the second excited state, whereas the ET-HisH+ state was the first excited state. For these two models, we computed the excitation energies with the CASSCF(2,3)3 method considering both excited states. Second, we computed five electronic states with the CASSCF(2,5)5 method considering the two lowestenergy LE states of flavin, the ET-3′T and ET-5′T states and the closed-shell ground state. The five-state calculations are significantly more time consuming than the two-state calculations, therefore we performed them only with the smaller (D,A) QM part. All CASSCF calculations were carried out using state-averaging with equal weights for all considered electronic states. The effect of dynamic electron correlation on the excitation energies was accounted for by using the multi-state extension of the second-order multiconfigurational perturbation theory (MS)CASPT2 method. The standard IPEA shift32 of 0.25 a.u. and an imaginary shift33 of 0.1 a.u. were used. Energy decomposition. The total energy of the molecular system consists of the quantummechanically computed QM/MM energy and the MM energy that includes all classically defined



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

10

Page 10 of 30

terms. The MM energy is the same for all electronic states and thus, does not influence the excitation energies. In the excitation energy analysis, we compare only the QM/MM energies. The QM/MM energy is the energy of the QM part that interacts, through the ESPF operator, with the surrounding electrostatic charges. We computed the QM/MM interaction energy by subtracting the QM/MM energy from the QM energy of the free QM part (no surrounding charges). The QM energy is computed at the same geometry as the QM/MM energy. The interaction energy has a positive sign and corresponds to the amount of energy by which the total QM energy is lowered by introducing the ESPF operator in the Hamiltonian. We compare the QM/MM interaction energy with the ESPF energy31 which is the energy of the electrostatic interactions of the MM-part RESP charges as defined by the AMBER94 force field and the QMpart ESPF charges fitted using the CASSCF electron density of each electronic state polarized by the embedding. The difference between the QM/MM interaction energy and the ESPF energy we consider as the polarization energy of the QM part by the MM part. We note that our definition of the polarization energy here is different from the one used in the original ESPF paper31, which considers the change of the ESPF interaction energy due to the QM-part polarization as the polarization energy. Results Optimized geometry of the active site. The active site of the DNA repair complex and the distribution of the Na+ counterions with respect to the DNA backbone are shown in Figure 1. The optimized geometries of the flavin hydroquinone anion and (6-4) photoproduct in the HisH+ and His models are very similar (the bond distances are compared in Figure S1 in the Supporting Information). The major difference is found for the orientation of the O4′H group of the 5′base that interacts with His365 either as a hydrogen-bond donor or a hydrogen-bond acceptor. The


Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


11

respective hydrogen-bond distance is equal to 2.7 Å in both models, which is equivalent to the distance in the parent crystal structure. The previous studies employing classical moleculardynamic simulations13 and geometry optimization of molecular clusters11 indicated that the hydrogen bond formed by protonated His365H+ has a longer Nδ…O4′ distance than that of the neutral counterpart. Evidently, the constraints implicated in the QM/MM geometry optimization do not allow accounting for this difference.

Figure 1. Structure of the DNA-repair site. A. In the HisH+ model, we indicate the electrondonor lumiflavin (LFH-), (6-4) photoproduct (3′T and 5′T), histidine side chain (HisH+) and the Na+ counterions (by their distances form the phosphorous atom in the circle). B. In the (D,A,H)



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

12

QM part, we indicate the O4′H group of the photoproduct, which is transferred from 5′T to 3′T in the repair course9. Excitation energies. All together we computed six electronic states of the repair active site. The MOs of the complete active space are shown in Figure 2. The ground state has a closed-shell structure with the doubly-occupied HOMO of flavin. In all excited states, the singly-occupied HOMO of flavin is combined with one of the singly-occupied LUMOs. The LE1 and LE2 states of flavin have singly-occupied LUMO and LUMO+1 of flavin; the ET-3′T, ET-5′T and ETHisH+ states have singly-occupied LUMO of the 3′T base, LUMO of the 5′T base and LUMO of the protonated HisH+, respectively.

Figure 2. The CASSCF-optimized MOs of the complete active space. The computed excitation energies are presented in Figure 3. Unless otherwise stated, we discuss the results obtained with the (MS)CASPT2 method. Computations of the closed-shell ground state and the ET-3′T state with the CASSCF(2,2)2 method provide the simplest model of


Page 12 of 30

Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


13

electron transfer in the photolyase-DNA complex. The ET-3′T excitation energy decreases if the histidine is protonated as presented in Figure 3A. The magnitude of the decrease is larger if the histidine side chain is included in the QM part. In the (D,A,H) QM models with protonated and neutral histidine, the excitation energies differ by 1.13 eV. Adding the MM part “compensates” the effect and the energy difference decreases to 0.57 eV. If the histidine side chain is included in the MM-part, the energy difference further reduces to 0.25 eV.

The CASSCF method

systematically overestimates the excitation energy by about 0.6 eV compared to the (MS)CASPT2 method. However, the parallel lines in Figures 3A and 3B demonstrate that the environmental effects do not differ to any appreciable extent with the CASSCF and CASPT2 methods revealing that the effects of dynamic electron correlation and the environment are largely additive for the considered cases.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

14

Figure 3. Excitation energies of the active site. A. The ET-3′T excitation energy computed with the CASSCF(2,2)2-(MS)CASPT2 method; B. The ET-3′T excitation energy computed with the CASSCF(2,2)2-(MS)CASPT2 method and models in which the MM part was modified. The energies of the HisH+(D,A,H) models with MM-6.9 and MM-(2.7+7.7) were computed with the


Page 14 of 30

Page 15 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


15

CASSCF(2,3)3-(MS)CASPT2 method (see text); C. The excitation energies computed with the CASSCF(2,5)5-(MS)CASPT2 method. The computed electronic states are ET-3′T (black), ET5′T (red), LE1 (dark-blue), LE2 (light-blue). In A-C, the models used for calculations are HisH+(D,A) (squares), HisH+(D,A,H) (triangles), His(D,A) (circles), His(D,A,H) (pentagons). Filled symbols and solid lines indicate the (MS)CASPT2 energies, open symbols and dashed lines indicate the CASSCF energies. The excitation energies computed with the modified MM part, in which the negative charge of DNA is less screened, are presented in Figure 3B. A series of modifications was considered for the HisH+(D,A) model, whereas for the other models only the limiting case of two ions removed MM-(2.7+7.7) or MM-2Na+ was computed. In the modified models, the ET excitation energies increase which is consistent with the reduced electron affinity of the photoproduct. We computed the largest effect by removing two ions at distances 2.7 and 7.7 Å. Removing the 2.7-Å ion increases the energy by 1.0 eV, whereas removing the 7.7-Å ion – by 0.5 eV. Removing the two ions simultaneously increases the energy by 1.5 eV. The additive effect indicates the dominating role of electrostatic interactions. Ions located farther than 12 Å away from the active site cause only a small effect. Removing the entire water shell but keeping all counterions and the crystalstructure water molecules increases the energy by about 1 eV, which is equivalent to removing the closest counterion to the active site. From this result, it is obvious, that solvating DNA in a QM/MM model is important for evaluation of the ET energies. There is a systematic difference in the excitation energies of the HisH+ and His models independent of the MM-part modifications. The protonated histidine side chain included in the QM part is an electron acceptor that yields a low-lying ET state. This excited state becomes lower in energy than the ET-3′T state in the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

16

Page 16 of 30

modified models MM-6.9 and MM-2Na+. In these two models, the ET-HisH+ excitation energy (not shown in Figure 3B) is 1.66 and 2.03 eV, whereas the ET-3′T energy (shown in Figure 3B) is 2.47 and 3.34 eV. These energies were computed with the CASSCF(2,3)3-(MS)CASPT2 method; the three-state ET-3′T energy are compared to the two-state energy of the unmodified model in Figure 3B. Notably, the same change of the environment increases the ET-HisH+ energy by 0.37 eV whereas the ET-3′T energy by 0.97 eV. Hence, not surprisingly, the electronacceptor property of His365H+ is influenced by the DNA charge to a lesser extent than that of the 3′base. For further characterization of the DNA repair site, we need to consider other excited states such as the LE states of the flavin chromophore and the ET states of other electron acceptors, in particular of the 5′base. The LE states give rise to the flavin absorption and fluorescence spectrum. In contrast, the ET states are the “dark” states with no intensity in the absorption spectrum. In the photolyase, the electron transfer to the DNA substrate is primarily responsible for fluorescence quenching6: the flavin fluorescence lifetime is significantly reduced if the enzyme binds the photodamaged DNA. Because the calculations of the five electronic states are significantly more demanding, we considered only the smaller (D,A) models. The excitation energies are presented in Figure 3C. We note that the energies of the ET-3′T state obtained with the CASSCF(2,2)2-(MS)CASPT2 and CASSCF(2,5)5-(MS)CASPT2 methods are very similar. The two LE states of flavin correspond to the HOMO–LUMO and HOMO–(LUMO+1) ππ* transitions with the energies around 2.95 eV (420 nm) and 3.40 eV (365 nm), respectively. The oscillator strengths of these transitions are 0.02 and 0.52, respectively (all computed oscillator strength are collected in the Supporting Information), indicating that the lowest-energy LE1 state results in a weak absorption whereas the second LE2 state produces a strong band. This


Page 17 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


17

computational finding is in good agreement with the UV–vis spectrum of FADH- in photolyases6,34 that have a strong band with a maximum around 380 nm and a low-intensity shoulder at the longer wave-lengths. In photolyases, the weak LE1 transition is activated via energy transfer from the folate or deazaflavin antenna chromophore. Comparing the spectra computed for the protonated and neutral active sites, we find that the excitation energy of LE1 state hardly changes. At the same time, the energy of the LE2 state decreases by 0.17 eV if the histidine is protonated. Our previous calculations based on supermolecular cluster models, revealed a very similar trend – a decrease from 3.36 to 3.27 eV of the LE2 excitation energy and a slight increase from 2.59 to 2.63 eV of the LE1 energy upon histidine protonation11. (The cluster-model calculations in ref.11 were performed at the B3LYP-optimized geometry using the XMCQDPT2 method to compute the excitation energies.) Upon removing the Na+ counterions, the LE1 and LE2 excitation energies practically stay unchanged. Now we compare the ET states associated with the two bases of the (6-4) photoproduct. The oscillator strengths of the ET states are virtually zero, consistent with the role of these states in fluorescence quenching. The energy of the ET-5′T state is higher than that of the ET-3′T state. Protonation of histidine decreases both excitation energies: by 0.23 eV the energy of the ET-3′T state (similar to what we found with the CASSCF(2,2)2 calculations of the (D,A) models) and by 0.59 eV the energy of the ET-5′T state. A significantly larger effect on the ET-5′T-state energy is explained by the hydrogen bond that the 5′base forms with His365. Upon removing the two Na+ ions, both excitation energies increase significantly: of the 5′base by 1.25 eV independent from the histidine protonation state, whereas of the 3′base by 1.5 and 1.6 eV in the HisH+ and His models, respectively. Similar to the two-state calculations, we find that the energy shifts due



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

18

Page 18 of 30

to the protonation and reduced screening of the DNA charge are practically independent from each other. QM/MM interaction energies. The interaction energies of the QM part with the MM part in all considered electronic states are compared in Figure 4. The differences in the interaction energies of the ground and excited electronic states explain the differences in excitation energies computed with the QM/MM models as compared to the free QM part: If the ground state interacts stronger with the MM environment, the excitation energy increases; if the excited state interacts stronger – the excitation energy decreases. In our QM/MM models, the differences in the interaction energies have two contributions: the changes in the charge distribution and the polarization of the electronic states. The polarization of the environment is not considered in our model. As Figure 4 demonstrates, the interaction energy is essentially the ESPF energy, i.e. the interaction energy of the electrostatic charges. The differences in the ESPF energy also determine the trends in the excitation energies caused by the effect of the environment. The difference between the ESPF energy and the QM/MM interaction energy corresponds to the polarization energy of the QM part by the electrostatic charges of the MM part. The ESPF energies are always larger than the QM/MM interaction energies thus the wave-function polarization decreases the interaction energies of the active site with its environment. Despite the differences in the CASSCF- and CASPT2-computed electrostatic charges, the CASSCF- and CASPT2-computed ESPF energies are very similar in magnitude (data not shown). The smaller QM/MM interaction energies computed with the (MS)CASPT2 method compared to the CASSCF method demonstrate the increased wave-function polarization in the former case. In the models with the modified MM-part, the increase of the ET excitation energies is explained by the decreasing interaction energies of the ET states. An exception is the HisH+(D,A,H) model with


Page 19 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


19

the reversed trend: the ground state interacts stronger with the environment than the excited ET state; the interaction of the ground state is even further enhanced when the two Na+ ions are removed. Obviously, this trend originates from the interactions of the cationic HisH+ side chain with the electrostatic environment.

Figure 4. The QM/MM interaction energies of the HisH+ models (A), His models (B), HisH+ models with the modified MM part (C). The electronic states are the closed-shell ground state



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

20

(green), ET-3′T (black), ET-5′T (red), LE1 (dark-blue), LE2 (light-blue). Open bars indicate the ESPF interaction energies computed with the CASSCF method, light-colored filled bars indicate the CASSCF QM/MM interaction energies, dark-colored filled bars indicate the (MS)CASPT2 QM/MM interaction energies. The two- and five-bar diagrams represent the CASSCF(2,2)2(MS)CASPT2 and CASSCF(2,5)5-(MS)CASPT2 calculations, respectively. The analysis of the electrostatic charge distribution in the active site explains the differences in the interaction energies. Naturally, the most pronounced difference is found for the ET states in comparison with the ground state as demonstrated in Figure 5. Upon electron transfer, the electron acceptor gains a negative charge, whereas the isoalloxazine ring of the flavin semiquinone radical is neutral. We note a significant charge transfer from the 3′base to the 5′base in the ground and excited states.

Figure 5. Mulliken electrostatic charges of the 3′base (triangles) and the 5′base (squares) in the ground closed-shell (green), ET-3′T (black) and ET-5′T (red) electronic states computed with the CASSCF method. The two-color diagrams are computed with CASSCF(2,2)2; the three-color diagrams are computed with the CASSCF(2,5)5 method.


Page 20 of 30

Page 21 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


21

Dynamic electron correlation. The comparison of dynamic electron-correlation energies of the considered electronic states computed as a difference of the (MS)CASPT2 and CASSCF excitation energies is presented in Figure 6. The differential dynamic electron-correlation energy has a tendency to increase with the increase of the excitation energy. Interestingly, the two-state calculations for the ET 3’T state are characterized by a smaller dynamic-correlation energy correction than the respective five-state calculations. This difference is explained by the differences in the respective CASSCF excitation energies: the ET-3′T excitation energy computed with the CASSCF(2,2)2 method is smaller than that computed with CASSCF(2,5)5, but after the (MS)CASPT2 correction very similar excitation energies are obtained with both methods. In Figure 6, two trends can be distinguished: one for the ET states and the other for the LE states.

Figure 6. Differential dynamic electron-correlation energy computed as a difference between the (MS)CASPT2 and CASSCF excitation energies. The electronic states are ET-3′T (black), ET5′T (red), LE1 (dark-blue), LE2 (light-blue). The models used for calculations are HisH+(D,A) (squares), HisH+(D,A,H) (triangles), His(D,A) (circles), His(D,A,H) (pentagons). “5-state” and “2-state”

indicate

the

CASSCF(2,5)5-(MS)CASPT2

and

calculations, respectively.


CASSCF(2,2)2-(MS)CASPT2


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

22

Page 22 of 30

Discussion and Conclusions We computed excitation energies of the DNA repair active site in the (6-4) photolyase-DNA macromolecular complex employing the hybrid QM/MM approach with the goal to study the effects of the environment on photoinduced electron transfer implicated in DNA repair. In particular, we computed the energies of the two LE states of the anionic flavin hydroquinone and the ET states of three electron acceptors: the two nucleobases of the photoproduct and the protonated histidine His365H+. These acceptors contribute to the quenching of flavin fluorescence in the photolyase-DNA complex. To compute on equal footing the LE and ET states we include the electron donor and the electron acceptor in the QM part. Two types of the environmental effects are of our interest: i) protonation of the histidine in the enzymatic active site which we evaluate by comparing the “protonated” and “neutral” active site models and ii) the effect of the DNA negative charge which we probed by altering the coordinates of the Na+ counterions in the MM part. We also analyzed how variations in the quantum-mechanical description such as including the dynamic electron correlation and considering different number of electronic states influence the excitation energies. The dynamic electron-correlation correction decreases the excitation energies with differential effects. Nevertheless, the effects of the environment on the excitation energy have a similar magnitude with the CASSCF and CASPT2 methods. The vertical excitation energies of the LE states are in good agreement with the excitation spectrum of the reduced flavin in the photolyase enzyme6,34 and also with the energies previously computed by us with the supermolecular cluster models11. Comparison of the cluster models and the QM/MM models is presented in Figure 7A. Among all considered states, the LE1 state of flavin is least affected by interactions with the ionic groups. The excitation energy of the LE2


Page 23 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


23

state slightly decreases from 3.4 to 3.2 eV upon histidine protonation. The DNA negative charge has a negligible effect on the LE excitation energies. In contrast, the excitation energies of the ET states are very sensitive to the presence of ionic groups. Protonation of the histidine leads to a significant stabilization of the ET states, especially in the models including histidine in the QM part. The DNA negative charge has an opposite effect, significantly increasing the ET excitation energies. We note that the cluster models predict smaller ET excitation energies than the QM/MM models. Lowering the ET-3′T energy is also observed upon extending the QM part by including the histidine side chain. This finding indicates that intermolecular interactions other than the electrostatic interactions, i.e. mutual polarization of the chromophore and environment, chargetransfer and dispersion that are not included in the QM/MM model, contribute to stabilization of the ET states. To account for these effects we need to extend the QM part in the QM/MM models, which is currently prohibitive due to the high computational cost. Another way of improving the QM/MM description is including the polarization of the environment (recently reviewed in ref.35). Accounting for the environment polarization may compensate the electrostatic-induced excitation-energy shifts as it is demonstrated for the tryptophan chromophore embedded in the human serum albumin protein35.Our cluster and QM/MM models predict energies of the ET states to be within a rather large energy interval – the lower bound is provided by our cluster models and the upper bound is provided by our QM/MM-Na+ model as shown in Figure 7A. We note that the differences in the ET excitation energies due to the variations of the molecular models are very large. Obviously, neither cluster nor QM/MM models represent an optimal choice. In fact, the choice of the model rather than the choice of the QM method may determine the accuracy of computations. Finding a balance between these two



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

24

Page 24 of 30

computational aspects represents, in our opinion, the major challenge faced in quantummechanical computations of the electron-transfer energies. The dependence of the photoproduct’s ET energies on the screening of the DNA negative charge points towards a potential drastic difference in the intrinsic properties of the nucleobases and the properties of DNA. To the best of our knowledge, the exciting picture of the DNA properties does not back up such a conjecture in general. In the context of the DNA-repair calculations, this finding indicates that the QM/MM models may predict varying fluorescence lifetimes, electron-transfer rates and radical-recombination rates depending on how the DNA substrate is treated. As depicted in Figure 7B, models with the less-screened DNA, for instance, the one employed in the study of ref.14, should predict longer flavin fluorescence lifetime and higher-energy radical pairs which undergo recombination more readily, whereas models with the more-screened DNA should predict shorter-lived fluorescence and more stable radical pairs. An important factor in DNA screening is the solvating water-shell, which is not present in the crystal structure but should be introduced in the QM/MM model. To obtain more reliable ET energy estimates we should study the counterion distribution with the classical molecular dynamics and identify the putative binding sites close to the active site (within about 12 Å from the phosphorous atom) in the photolyase-DNA complex.


Page 25 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


25

Figure 7. Electron and proton transfer energies. A. Comparison of the excitation energies obtained with the QM/MM models and cluster models11; B. The change of the ET-state energy along the electron-transfer pathway leads to changes in the lifetimes of the excited flavin and radical pair. The higher ET energies result from the decreased screening of the DNA negative charge; the lower ET energies result from the protonation of the active site, screening (or omitting) of the DNA charge and also from extending the QM part; C. Energies along the proton-coupled electron transfer reaction coordinate. Electron transfer from FADH- to 3′base; proton transfer from His365H+ to 3′base; the QM/MM energies of the protonated 3′base model are determined by energy extrapolation on the basis of the cluster-model results11. We extended the QM part by including the histidine side chain in order to compute protonation of the (6-4) photoproduct by the histidine in our further studies. This protonation, following photoinduced electron transfer, is considered by several authors to be a critical step in the DNA



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

26

Page 26 of 30

repair mechanism12,14,15. The resulting electron and proton transfer model is computationally very demanding to use for the QM/MM geometry optimization with the CASSCF method or for computing the higher-energy excited states. Nevertheless, we were able to compare the excitation energy of the flavin-photoproduct ET-3′T state computed with the models containing the neutral or protonated His365 side chain in the QM part. In Figure 7C, we show, according to our previous calculations11, that proton and electron transfer lowers the energy of the ET-3′T state, which becomes the ground state. The excited-state energy decrease upon proton transfer is 2.7 eV. This decrease significantly exceeds the increase of the ET-3′T-state excitation energy in the QM/MM-Na+ models, in which electron transfer is hindered. Therefore, we can expect the ground-state proton coupled electron transfer to take place even under the unfavorable conditions of the reduced DNA screening. This preliminary result suggests that the deactivation of the enzyme via the ground-state proton-coupled electron transfer is a likely process. In experiments, this enzyme’s deactivation is probably observed upon lowering the pH from 9 to 74. Thus, the proposed catalytic role of protonation must be carefully reevaluated not only by further computational studies but also by specifically designed experiments. We also found that protonated His365H+ is a plausible electron acceptor, which may contribute to the quenching of the FADH- fluorescence at the proton concentration lower than the pKa of His365. For the wild-type (6-4) photolyase and a series of its His365 mutants, electron transfer dynamics have been characterized by means of ultrafast spectroscopy6. In these proteins without the DNA-substrate bound, the lifetime of flavin fluorescence ranges from 4 to 1.7 ns. Upon DNA-substrate binding, the fluorescence lifetime decreases at least by an order of magnitude revealing that the ET states of the photoproduct are involved in fluorescence quenching. Our calculations presented here reveal that the ET-3′T and ET-5′T states are the electronic states


Page 27 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


27

contributing to the quenching process. Only in the wild-type protein (containing His365) the electron-transfer process with the time constant of 225 ps results in DNA repair. In the mutated proteins, the time constants vary from 281 to 147 ps but no repair takes place. In the original paper6, this observation was interpreted as evidence for the presence of the protonated His365H+ side chain. However, there is no correlation between the determined electron-transfer rates and the electrostatic charge or polarity of the mutated side chain. This observation rather agrees with the notion supported by our studies that the repair-competent active site contains the neutral His365 side chain. Interestingly, in the photodynamic experiments, protonated cationic lysine is similar to deprotonated anionic aspartate and neutral alanine (time constants 281, 278 and 274 ps, respectively)6 indicating that the presence of the charged side chain is unlikely to play a role in the observed rates. The wild-type photolyase is similar to the mutants with neutral residues methionine and tyrosine (time constants 225, 232 and 249 ps, respectively)6. The most shortlived fluorescence (time constant 147 ps) was found in the asparagine mutant6. These intriguing observations await an explanation. For the development of adequate kinetic models, computational models that evaluate electron-transfer rates quantum-mechanically and also combine the account of various relevant effects, for instance protein and DNA molecular dynamics, are indispensible despite all technical challenges and complexity. Our work constitutes the first step towards this goal. Acknowledgments We are very grateful to Chris Roome for the excellent assistance in high-performance computing, to Nicolas Ferré (Aix Marseille Université) for computational advice, to Ilme Schlichting for collaboration and support. We acknowledge financial support from the MPG Minerva program (T.D.), and BIOMS-Heidelberg (A.R.M.S.).



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

28

Page 28 of 30

Supporting Information Available Selected distances in the optimized geometries are compared in Figure S1. All computed energies are collected in Tables S1 and S2, the Mulliken charges – Tables S3 and S4, the Cartesian coordinates of the optimized geometries of the QM parts – Tables S5 and S6. This information is available free of charge via the Internet at http://pubs.acs.org. References: (1) Weber, S. Biochim. Biophys. Acta 2005, 1707, 1-23. (2) Sancar, A. J. Biolog. Chem. 2008, 283, 32153-32157. (3) Muller, M.; Carell, T. Curr. Opin. Struct. Biol. 2009, 19, 277-285. (4) Hitomi, K.; Nakamura, H.; Kim, S. T.; Mizukoshi, T.; Ishikawa, T.; Iwai, S.; Todo, T. J. Biol. Chem. 2001, 276, 10103-10109. (5) Schleicher, E.; Hitomi, K.; Kay, C. W.; Getzoff, E. D.; Todo, T.; Weber, S. J. Biol. Chem. 2007, 282, 4738-4747. (6) Li, J.; Liu, Z. Y.; Tan, C.; Guo, X. M.; Wang, L. J.; Sancar, A.; Zhong, D. P. Nature 2010, 466, 887-890. (7) Maul, M. J.; Barends, T. R. M.; Glas, A. F.; Cryle, M. J.; Domratcheva, T.; Schneider, S.; Schlichting, I.; Carell, T. Angew. Chem. Int. Ed. 2008, 47, 10076-10080. (8) Fingerhut, B. P.; Heil, K.; Kaya, E.; Oesterling, S.; de Vivie-Riedle, R.; Carell, T. Chem. Sci. 2012, 3, 1794-1797. (9) Domratcheva, T.; Schlichting, I. J. Am. Chem. Soc. 2009, 131, 17793-17799. (10) Harbach, P. H. P.; Borowka, J.; Bohnwagner, M. V.; Dreuw, A. J. Phys. Chem. Lett. 2010, 1, 2556-2560. (11) Domratcheva, T. J. Am. Chem. Soc. 2011, 133, 18172-18182. (12) Faraji, S.; Dreuw, A. J. Phys. Chem. Lett. 2012, 3, 227-230. (13) Condic-Jurkic, K.; Smith, A. S.; Zipse, H.; Smith, D. M. J. Chem. Theory Comput. 2012, 8, 1078-1091. (14) Sadeghian, K.; Bocola, M.; Merz, T.; Schutz, M. J. Am. Chem. Soc. 2010, 132, 16285-16295. (15) Ai, Y. J.; Liao, R. Z.; Chen, S. L.; Hua, W. J.; Fang, W. H.; Luo, Y. J. Phys. Chem. B 2011, 115, 10976-10982. (16) Ferre, N.; Angyan, J. G. Chem. Phys. Lett. 2002, 356, 331-339. (17) Aquilante, F.; De Vico, L.; Ferre, N.; Ghigo, G.; Malmqvist, P. A.; Neogrady, P.; Pedersen, T. B.; Pitonak, M.; Reiher, M.; Roos, B. O.; Serrano-Andres, L.; Urban, M.; Veryazov, V.; Lindh, R. J. Comput. Chem. 2010, 31, 224-247. (18) Karlstrom, G.; Lindh, R.; Malmqvist, P. A.; Roos, B. O.; Ryde, U.; Veryazov, V.; Widmark, P. O.; Cossi, M.; Schimmelpfennig, B.; Neogrady, P.; Seijo, L. Comput. Mater. Sci. 2003, 28, 222-239. (19) Ponder, J. W. TINKER: Software tools for molecular design, version 5.1, 2010, http://dasher.wustl.edu/tinker. Accessed date - August 2013.


Page 29 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


29

(20) Roos, B. O.; Andersson, K.; Fulscher, M. P.; Malmqvist, P. A.; Serrano Andres, L.; Pierloot, K.; Merchan, M. Adv. Chem. Phys. 1996, 93, 219-331. (21) Andersson, K.; Malmqvist, P. A.; Roos, B. O. J. Chem. Phys. 1992, 96, 12181226. (22) Finley, J.; Malmqvist, P. A.; Roos, B. O.; Serrano-Andres, L. Chem. Phys. Lett. 1998, 288, 299-306. (23) Harihara, P; Pople, J. A. Theor. Chim. Acta 1973, 28, 213-222. (24) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1996, 118, 2309-2309. (25) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926-935. (26) Martinez, L.; Andrade, R.; Birgin, E. G.; Martinez, J. M. J. Comput. Chem. 2009, 30, 2157-2164. (27) Masson, F.; Laino, T.; Rothlisberger, U.; Hutter, J. Chem. Phys. Chem. 2009, 10, 400-410. (28) Spector, T. I.; Cheatham, T. E.; Kollman, P. A. J. Am. Chem. Soc. 1997, 119, 7095-7104. (29) Humbel, S.; Sieber, S.; Morokuma, K. J. Chem. Phys. 1996, 105, 1959-1967. (30) Sherwood, P.; de Vries, A. H.; Collins, S. J.; Greatbanks, S. P.; Burton, N. A.; Vincent, M. A.; Hillier, I. H. Faraday Discuss. 1997, 106, 79-92. (31) Melaccio, F.; Olivucci, M.; Lindh, R.; Ferre, N. Intern. J. Quant. Chem. 2011, 111, 3339-3346. (32) Ghigo, G.; Roos, B. O.; Malmqvist, P. A. Chem. Phys. Lett. 2004, 396, 142-149. (33) Forsberg, N.; Malmqvist, P. A. Chem. Phys. Lett. 1997, 274, 196-204. (34) Kao, Y. T.; Saxena, C.; Wang, L. J.; Sancar, A.; Zhong, D. P. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 16128-16132. (35) Monari, A.; Rivail, J.-L.; Assfeld, X. Acc. Chem. Res. 2013, 46, 596-603.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

For Table of Contents Use Only


30

Page 30 of 30

Challenges in Computing Electron-Transfer Energies of DNA Repair

Recommend Documents